Launching Project Genie for Creating Interactive Worlds

In August, we introduced Genie 3 — a universal world model capable of generating a variety of interactive environments. Even in its early version, trusted testers were able to create an impressive array of engaging worlds and experiences, discovering entirely new ways to utilize this tool.

Starting today, we are granting access to Project Genie for Google AI Ultra subscribers in the U.S. (18+). This experimental research prototype allows users to create, explore, and modify their own interactive worlds.

The world model simulates environmental dynamics, predicting how it will evolve and how actions affect it. While Google DeepMind has experience working with agents in specific environments, such as chess or Go, creating AGI requires systems capable of navigating the diversity of the real world.

To tackle this challenge and support our mission of creating AGI, we developed Genie 3. Unlike static 3D snapshots, Genie 3 generates a path forward in real-time as you move and interact with the world. It simulates physics and interactions for dynamic worlds, and its groundbreaking consistency allows for modeling any real-life scenarios — from robotics and animation simulation to exploring locations and historical events.

Building on our research with trusted testers from various industries, we are taking the next step with the experimental research prototype: Project Genie.

Project Genie is a web application prototype powered by Genie 3, Nano Banana Pro, and Gemini, allowing users to experiment with immersive experiences of our world model directly.